update transcript schema

Author: yujonglee

apps/desktop/src/components/main/body/sessions/floating/listen.tsx

@@ -174,22 +174,42 @@ function useStartSession(sessionId: string) {
 
 function useAppendTranscript(sessionId: string) {
   const store = persisted.UI.useStore(persisted.STORE_ID);
+  const transcriptIds = persisted.UI.useSliceRowIds(
+    persisted.INDEXES.transcriptsBySession,
+    sessionId,
+    persisted.STORE_ID,
+  );
 
   const handler = useCallback((res: StreamResponse) => {
     if (store && res.type === "Results") {
-      res.channel.alternatives[0].words.forEach((w) => {
-        store.setRow("words", id(), {
+      let transcriptId = transcriptIds?.[0];
+
+      if (!transcriptId) {
+        transcriptId = id();
+        store.setRow("transcripts", transcriptId, {
           session_id: sessionId,
+          user_id: "",
+          created_at: new Date().toISOString(),
+        });
+      }
+
+      const { channel: { alternatives: [{ words }] }, channel_index } = res;
+
+      words.forEach((w) => {
+        const word: persisted.Word = {
+          transcript_id: transcriptId,
           text: w.word,
           start_ms: Math.round(w.start * 1000),
           end_ms: Math.round(w.end * 1000),
-          speaker: w.speaker?.toString() ?? undefined,
+          channel: channel_index[0],
           user_id: "",
           created_at: new Date().toISOString(),
-        });
+        };
+
+        store.setRow("words", id(), word);
       });
     }
-  }, [store, sessionId]);
+  }, [store, sessionId, transcriptIds]);
 
   return handler;
 }

apps/desktop/src/components/main/body/sessions/note-input/transcript.tsx

@@ -4,15 +4,24 @@ import * as persisted from "../../../../../store/tinybase/persisted";
 export function TranscriptView({ sessionId }: { sessionId: string }) {
   const store = persisted.UI.useStore(persisted.STORE_ID);
 
+  const transcriptIds = persisted.UI.useSliceRowIds(
+    persisted.INDEXES.transcriptsBySession,
+    sessionId,
+    persisted.STORE_ID,
+  );
+  const transcriptId = transcriptIds?.[0];
+
   const QUERY = `${sessionId}_words`;
   const QUERIES = persisted.UI.useCreateQueries(
     store,
     (store) =>
       createQueries(store).setQueryDefinition(QUERY, "words", ({ select, where }) => {
         select("text");
-        where("session_id", sessionId);
+        if (transcriptId) {
+          where("transcript_id", transcriptId);
+        }
       }),
-    [sessionId],
+    [sessionId, transcriptId],
   );
 
   const words = persisted.UI.useResultTable(QUERY, QUERIES);

apps/desktop/src/devtool/seed/shared.ts

@@ -15,6 +15,7 @@ import type {
   Tag,
   TemplateSection,
   TemplateStorage,
+  Transcript,
   Word,
 } from "../../store/tinybase/persisted";
 import { id } from "../../utils";
@@ -137,8 +138,7 @@ export const generateEnhancedMarkdown = () => {
 
 export const generateTranscript = () => {
   const wordCount = faker.number.int({ min: 50, max: 200 });
-  const words: Array<{ speaker: string; text: string; start_ms: number; end_ms: number }> = [];
-  const speakers = ["Speaker 1", "Speaker 2"];
+  const words: Array<Word> = [];
 
   let currentTimeMs = 0;
 
@@ -147,7 +147,10 @@ export const generateTranscript = () => {
     const durationMs = faker.number.int({ min: 200, max: 800 });
 
     words.push({
-      speaker: faker.helpers.arrayElement(speakers),
+      user_id: USER_ID,
+      created_at: faker.date.recent({ days: 30 }).toISOString(),
+      transcript_id: id(),
+      channel: 0,
       text: word,
       start_ms: currentTimeMs,
       end_ms: currentTimeMs + durationMs,
@@ -382,6 +385,7 @@ export const generateMockData = (config: MockConfig) => {
   const calendars: Record<string, Calendar> = {};
   const folders: Record<string, Folder> = {};
   const sessions: Record<string, SessionStorage> = {};
+  const transcripts: Record<string, Transcript> = {};
   const words: Record<string, Word> = {};
   const events: Record<string, Event> = {};
   const mapping_session_participant: Record<string, mappingSessionParticipant> = {};
@@ -494,16 +498,23 @@ export const generateMockData = (config: MockConfig) => {
       sessions[session.id] = session.data;
       sessionIds.push(session.id);
 
+      const transcriptId = id();
+      transcripts[transcriptId] = {
+        user_id: USER_ID,
+        session_id: session.id,
+        created_at: faker.date.recent({ days: 30 }).toISOString(),
+      };
+
       const transcript = generateTranscript();
       transcript.words.forEach((word) => {
         const wordId = id();
         words[wordId] = {
           user_id: USER_ID,
-          session_id: session.id,
+          transcript_id: transcriptId,
           text: word.text,
           start_ms: word.start_ms,
           end_ms: word.end_ms,
-          speaker: word.speaker || undefined,
+          channel: word.channel,
           created_at: faker.date.recent({ days: 30 }).toISOString(),
         };
       });
@@ -553,6 +564,7 @@ export const generateMockData = (config: MockConfig) => {
     calendars,
     folders,
     sessions,
+    transcripts,
     words,
     events,
     mapping_session_participant,

apps/desktop/src/store/tinybase/internal.ts

@@ -3,6 +3,7 @@ import { createMergeableStore, createQueries, type MergeableStore, type TablesSc
 import { z } from "zod";
 
 import { createBroadcastChannelSynchronizer } from "tinybase/synchronizers/synchronizer-broadcast-channel/with-schemas";
+import { DEFAULT_USER_ID } from "../../utils";
 import { createLocalPersister } from "./localPersister";
 import { type InferTinyBaseSchema, jsonObject, type ToStorageType } from "./shared";
 
@@ -98,8 +99,7 @@ export const createStore = () => {
 export const useStore = () => {
   const store = useCreateMergeableStore(() => createStore());
 
-  // TODO
-  store.setValue("user_id", "4c2c0e44-f674-4c67-87d0-00bcfb78dc8a");
+  store.setValue("user_id", DEFAULT_USER_ID);
 
   const synchronizer = useCreateSynchronizer(
     store,

apps/desktop/src/store/tinybase/persisted.ts

@@ -27,6 +27,7 @@ import {
   TABLE_SESSIONS,
   tagSchema as baseTagSchema,
   templateSchema as baseTemplateSchema,
+  transcriptSchema as baseTranscriptSchema,
   wordSchema,
 } from "@hypr/db";
 import { createBroadcastChannelSynchronizer } from "tinybase/synchronizers/synchronizer-broadcast-channel/with-schemas";
@@ -67,6 +68,10 @@ export const sessionSchema = baseSessionSchema.omit({ id: true }).extend({
   folder_id: z.preprocess(val => val ?? undefined, z.string().optional()),
 });
 
+export const transcriptSchema = baseTranscriptSchema.omit({ id: true }).extend({
+  created_at: z.string(),
+});
+
 export const mappingSessionParticipantSchema = baseMappingSessionParticipantSchema.omit({ id: true }).extend({
   created_at: z.string(),
 });
@@ -99,6 +104,7 @@ export const chatMessageSchema = baseChatMessageSchema.omit({ id: true }).extend
 export const wordSchemaOverride = wordSchema.omit({ id: true }).extend({
   created_at: z.string(),
   speaker: z.preprocess(val => val ?? undefined, z.string().optional()),
+  transcript_id: z.string(),
 });
 
 export type Human = z.infer<typeof humanSchema>;
@@ -107,6 +113,7 @@ export type Calendar = z.infer<typeof calendarSchema>;
 export type Organization = z.infer<typeof organizationSchema>;
 export type Folder = z.infer<typeof folderSchema>;
 export type Session = z.infer<typeof sessionSchema>;
+export type Transcript = z.infer<typeof transcriptSchema>;
 export type Word = z.infer<typeof wordSchemaOverride>;
 export type mappingSessionParticipant = z.infer<typeof mappingSessionParticipantSchema>;
 export type Tag = z.infer<typeof tagSchema>;
@@ -117,6 +124,7 @@ export type ChatGroup = z.infer<typeof chatGroupSchema>;
 export type ChatMessage = z.infer<typeof chatMessageSchema>;
 
 export type SessionStorage = ToStorageType<typeof sessionSchema>;
+export type TranscriptStorage = ToStorageType<typeof transcriptSchema>;
 export type TemplateStorage = ToStorageType<typeof templateSchema>;
 export type ChatMessageStorage = ToStorageType<typeof chatMessageSchema>;
 
@@ -138,14 +146,19 @@ const SCHEMA = {
       raw_md: { type: "string" },
       enhanced_md: { type: "string" },
     } satisfies InferTinyBaseSchema<typeof sessionSchema>,
+    transcripts: {
+      user_id: { type: "string" },
+      created_at: { type: "string" },
+      session_id: { type: "string" },
+    } satisfies InferTinyBaseSchema<typeof transcriptSchema>,
     words: {
       user_id: { type: "string" },
       created_at: { type: "string" },
       text: { type: "string" },
-      session_id: { type: "string" },
+      transcript_id: { type: "string" },
       start_ms: { type: "number" },
       end_ms: { type: "number" },
-      speaker: { type: "string" },
+      channel: { type: "number" },
     } satisfies InferTinyBaseSchema<typeof wordSchema>,
     humans: {
       user_id: { type: "string" },
@@ -318,6 +331,18 @@ export const StoreComponent = () => {
           "folders",
           "parent_folder_id",
         )
+        .setRelationshipDefinition(
+          "transcriptToSession",
+          "transcripts",
+          "sessions",
+          "session_id",
+        )
+        .setRelationshipDefinition(
+          "wordToTranscript",
+          "words",
+          "transcripts",
+          "transcript_id",
+        )
         .setRelationshipDefinition(
           "sessionParticipantToHuman",
           "mapping_session_participant",
@@ -436,6 +461,8 @@ export const StoreComponent = () => {
       .setIndexDefinition(INDEXES.sessionsByHuman, "mapping_session_participant", "human_id")
       .setIndexDefinition(INDEXES.foldersByParent, "folders", "parent_folder_id", "name")
       .setIndexDefinition(INDEXES.sessionsByFolder, "sessions", "folder_id", "created_at")
+      .setIndexDefinition(INDEXES.transcriptsBySession, "transcripts", "session_id")
+      .setIndexDefinition(INDEXES.wordsByTranscript, "words", "transcript_id", "start_ms")
       .setIndexDefinition(INDEXES.eventsByCalendar, "events", "calendar_id", "started_at")
       .setIndexDefinition(
         INDEXES.eventsByDate,
@@ -529,6 +556,8 @@ export const INDEXES = {
   sessionParticipantsBySession: "sessionParticipantsBySession",
   foldersByParent: "foldersByParent",
   sessionsByFolder: "sessionsByFolder",
+  transcriptsBySession: "transcriptsBySession",
+  wordsByTranscript: "wordsByTranscript",
   eventsByCalendar: "eventsByCalendar",
   eventsByDate: "eventsByDate",
   sessionByDateWithoutEvent: "sessionByDateWithoutEvent",

apps/desktop/src/store/zustand/tabs/test-utils.ts

@@ -1,10 +1,9 @@
+import { id } from "../../../utils";
 import { type Tab, useTabs } from ".";
 
 type SessionTab = Extract<Tab, { type: "sessions" }>;
 type ContactsTab = Extract<Tab, { type: "contacts" }>;
 
-const id = () => crypto.randomUUID();
-
 type SessionOverrides = Partial<Omit<SessionTab, "type" | "state">> & {
   state?: Partial<SessionTab["state"]>;
 };

apps/desktop/src/utils.ts

@@ -0,0 +1,6 @@
+export * from "./timeline";
+
+export const id = () => crypto.randomUUID() as string;
+
+// https://www.rfc-editor.org/rfc/rfc4122#section-4.1.7
+export const DEFAULT_USER_ID = "00000000-0000-0000-0000-000000000000";

apps/desktop/src/utils/index.ts

@@ -0,0 +1,51 @@
+import { describe, expect, it } from "vitest";
+
+import { buildSegments } from "./transcript";
+
+describe("buildSegments", () => {
+  const WORD = (
+    {
+      text,
+      channel,
+      start_ms,
+      end_ms,
+    }: {
+      text: string;
+      channel: number;
+      start_ms: number;
+      end_ms: number;
+    },
+  ) => ({
+    user_id: "TODO",
+    created_at: "TODO",
+    transcript_id: "TODO",
+    text,
+    channel,
+    start_ms,
+    end_ms,
+  });
+
+  it.each([
+    {
+      description: "returns empty array for empty input",
+      words: [],
+      segments: [],
+    },
+    {
+      description: "creates single segment from single word",
+      words: [
+        WORD({ text: "hello", channel: 0, start_ms: 0, end_ms: 100 }),
+      ],
+      segments: [
+        { speaker: "Speaker 0", text: "hello" },
+      ],
+    },
+  ])("$description", ({ words, segments }) => {
+    const result = buildSegments({
+      words,
+      speakerFromChannel: (channel) => `Speaker ${channel}`,
+    });
+
+    expect(result).toEqual(segments);
+  });
+});

apps/desktop/src/utils/transcript.test.ts

@@ -0,0 +1,25 @@
+import type { Word } from "../store/tinybase/persisted";
+
+type Segment = {
+  text: string;
+  speaker: string;
+};
+
+export const buildSegments = (
+  {
+    speakerFromChannel,
+    words,
+  }: {
+    speakerFromChannel: (channel: number) => string;
+    words: Word[];
+  },
+): Segment[] => {
+  const segments: Segment[] = [];
+
+  for (const word of words) {
+    const speaker = speakerFromChannel(word.channel);
+    segments.push({ text: word.text, speaker });
+  }
+
+  return segments;
+};